On Sat, 5 Jan 2008, Stephen J. Turnbull wrote: I understand why they backport things, but do you know why they decide to link in the flyspell.el into both the emacs and xemacs trees, when both the emacs (emacs22-el and emacs22-common) and xemacs (xemacs21-basesupport and xemacs21-basesupport-el) packages have their own versions? I am a little confused. On a Centos 4 machine where I have not been able to break flyspell at all, the flyspell.el is dating from 2002. The same file, modulo a tiny patch involving - (if (fboundp 'about-xemacs) + (if (featurep 'xemacs) (sit-for flyspell-delay nil) (sit-for flyspell-delay 0 nil))) is in debian's xemacs21-basesupport package. The 2007 version in dictionaries-common is obviously quite a large patch. The 2007 version is the one that acts global instead of as buffer local. However, lastnight I found that the 2002 version on debian, had a fault with button-2 - this bug, by the looks of things: http://groups.google.co.uk/group/comp.emacs.xemacs/browse_thread/thread/1... The reporter obviously never solved that bug, instead getting a self-consistent version of the sumo instead of debian's version. Despite being basically the same file in Centos4, the button-2 fault does not appear in it. So I installed flyspell.el (v 1.7n) from http://www-sop.inria.fr/mimosa/Manuel.Serrano/flyspell/flyspell.html and it seems to be working perfectly on debian, it having overridden the version in xemacs21-basesupport, and of course the one in dictionaries-common. That's where I am now -- using upstream's 1.7n version on debian, and am not going to touch the Centos4 machine until it breaks:) -- Tim Connors _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta